Grid Workflow Scheduling based on Incomplete Information

Size: px
Start display at page:

Download "Grid Workflow Scheduling based on Incomplete Information"

Transcription

1 Grid Workflow Scheduling based on Incomplete Information vorgelegt von Diplom-Informatiker Jörg Schneider aus Berlin von der Fakultät IV - Elektrotechnik und Informatik der Technischen Universität Berlin zur Erlangung des akademischen Grades Doktor der Ingenieurwissenschaften Dr.-Ing. genehmigte Dissertation Promotionsausschuss: Vorsitzender: Prof. Dr. Peter Pepper Gutachter: Prof. Dr. Hans-Ulrich Heiß Gutachter: Prof. Dr. Odej Kao Gutachter: Prof. Dr. César de Rose Tag der wissenschaftlichen Aussprache: Berlin 2010 D 83

2

3 Acknowledgments I would like to thank especially my supervisor Prof. Dr. Hans-Ulrich Heiß for the opportunity to work with him and to write this thesis. I really appreciated the independence and freedom working in his research group. I also thank Prof. Dr. César de Rose, PUCRS Porto Alegre, and Prof. Dr. Odej Kao for serving as reviewer for this thesis. Special thanks to my colleagues and the students I worked with for the discussions, motivation, and support. In particular, I like to thank Sebastian Boelter, Dr. Lars-Olof Burchard, Jörg Decker, Stanimir Dragiev, Tiago Ferreto, Julius Gehr, Barry Linnert, Arthur Lochstampfer, and Robert Lubkoll. 3

4

5 Contents 1 Introduction 9 2 Coupling Resources in the Grid How to build up a Grid? Timing: Advance Reservation Formalization Manage advance reservations Measuring Fragmentation Typical properties of advance reservation systems Grid workflow Example Grid Workflows Formalization Heterogeneous Grid environment and moldable jobs Generating workloads containing Grid workflows The Virtual Resource Manager - A Grid Resource Manager Grid Workflow Languages Workflow Language and Representations Taxonomy of Grid Workflow Languages Related Work Language Properties Representation Properties Examples of Grid Workflow Languages YAWL - Yet another workflow language AGWL and CGWL GJobDL GridAnt Unicore VRM Evaluation of the Taxonomy Fast Admission of Grid Workflows What is a Good Online Scheduler?

6 Contents 4.2 Admission Protocol Communication between a Grid Broker and Resources Communication between a User and Grid Brokers Level of Commitment vs. Information Hiding Related Algorithms Greedy Algorithms Using single job admissions List Scheduler Reduce makespan: HEFTSync HEFTSyncBT: Backtracking Extension HLST - Applying the HEFT Algorithm Backwards Optimization Algorithms Defining a Constraint Satisfaction Problem Optimization Function Heterogeneous Grid environments and moldable jobs Evaluation Simulation Setup Influence of the proposed admission Protocol Online scheduling algorithms Re-Optimizing the Schedule Related work What is a Good Offline Scheduler? Grid Level vs. Resource Level Grid initiated optimization Extending the admission protocol Optimization on Demand Algorithms Reduce the Search Space Local initiated optimization Extending the Admission protocol Optimization Scheme Local only optimization Rerouting of network reservations Evaluation Optimization at Grid Level Network rerouting Handling Resource Outages by Re-Scheduling Related Work Failure Model and Objectives

7 Contents 6.3 Defining a Rescue horizon Simple reference models Objectives for the rescue horizon Load based horizon definition Feedback based horizon definition Evaluation Conclusion and Future Work 119 Previously published work 123 Bibliography 125 7

8

9 1 Introduction fast specifies admission input optimize schedule handle resource downtime While the processing power of single computers is continuously rising, there is still a need for supercomputers as the size of the problems to be solved increases, too. Most supercomputers are now connected to the Internet making them easily accessible from all over the world and scientists even have a choice of using multiple supercomputers of different providers. These providers are not only classical compute centers, but also companies and scientific institutions owning high performance computers and selling unused capacity. In order to use the potential of all the available computing power, it is necessary to know the hardware and software equipment as well as administrative regulation of each supercomputer. To solve this problem, the concept of Grid computing was developed. The goal is to provide high performance computing power in a reliable, standardized, and easy to use way, similar to electrical power in the energy grid. In the same way as the energy grid, the computational Grid spans multiple loosely coupled organizations providing services using a wide range of hardware. The computational Grid even includes non-uniform services, i.e., also non-computer resources like network bandwidth, storage capacity, or rare scientific instruments. A similar concept is the Cloud computing. The Cloud enables the access to diverse resources, too. However, resources are booked for an undefined time period in the Cloud and provide capacity for repeated service usages, e.g., by webservice calls. In the definition used in this thesis, Grid jobs are executed only once and report then the results back to the submitter. Despite a lot of similarities between the Cloud and the Grid, the techniques developed in this thesis focus on Grid environments with scientific applications. However, some results can be adapted to be used in the Cloud. To use the variety of services and resources in a coordinated manner, the user needs a more complex interface than the power plug in the energy grid. Grid workflows have been established as a way to define complex Grid applications as a set of jobs and to express the interdependencies between them. The user models his application as a Grid workflow and submits it to a Grid broker. The Grid broker then negotiates with all connected resource provider allocations matching the resource and timing requirements of the user. The Grid broker can also guarantee a specific service level, e.g., that all involved resources provide security measures or that there will be enough capacity to complete the Grid workflow by a userspecified deadline. In this case the Grid broker also has to control whether all 9

10 1 Introduction fast admission specifies input optimize schedule handle resource downtime Figure 1.1: The whole lifecycle of a Grid workflow from the Grid brokers point of view. involved resources stick to the guaranteed service level. The Grid broker can allocate resources most efficiently being in control of every resource in the Grid with very detailed information about their state. However, in a widely distributed and multi-organizational setup this is rarely possible. Resource providers desire to stay in charge of their resources and also control which information is available about the resource and its state. Typical information to hide is the load of the resource. This is necessary for a variety of reasons, like preventing a competing company from monitoring the load and inferring about internal processes of the resource owner, e.g., the release cycles of future products. Thus, the Grid broker utilizing such resources has to make its decisions with as little information and control as possible. This thesis provides means for the Grid broker to handle Grid workflows in such an environment. There are four stages in the lifecycle of the Grid workflow from the Grid broker s point of view: specification, online scheduling, offline scheduling, and rescue scheduling. For all these four stages, the specific requirements are analyzed and techniques are presented to efficiently cope with them. In order to provide guarantees for the deadline and to support co-allocation the allocation of multiple resources for the very same time advance reservations have to be used. In contrast to the widely used queuing approach, an advance reservation enabled resource manager negotiates the actual execution time during 10

11 the admission of the job. Therefore, the resource manager has to keep track of all future reservations and their start time. The thesis starts with an overview and comparison of potential data structures to efficiently handle multi-unit schedules with advance reservation. As the first stage in the lifecycle, the Grid workflow has to be specified by the client. There are a large number of Grid workflow languages with a wide variety of features. To support the selection of the language best matching the specific requirements, a taxonomy of Grid workflow languages was developed. In Chapter 3 this taxonomy is used to compare a number of known Grid workflow description languages. This classification does not only give an overview on the Grid workflow languages currently most often used, but also shows that the taxonomy covers the parameter range very well without introducing redundant classes. The remaining three stages handle the actual allocation process of a Grid workflow. The first allocation happens, when the Grid workflow is submitted and a valid reservation for all parts of the Grid workflow has to be found to negotiate the service level agreement (SLA). This online scheduling problem, covered in Chapter 4, is meant to be fast and may return empty handed even if there are valid allocations. Existing algorithms for Grid workflow scheduling lack support for co-allocations and guaranteed deadlines. Additionally, most of these algorithms require a detailed view on the state of each resource in the Grid. To overcome these limitations, a newly developed admission protocol is introduced to efficiently negotiate the coordinated allocation without exchanging too much information. The protocol has three phases: First, the Grid broker receives non-binding reservation candidates after sending probe messages for each part of the Grid workflow, then the Grid broker preliminary reserves the set of reservation candidates best fitting the Grid workflow constraints, and, finally, it commits these reservations if the previous phase was successful for all jobs in the Grid workflow. Simulations showed that without the probe phase, significant more reservation requests were send to the resources during the allocation. Using the probe results of the admission protocol, Grid broker can employ two general strategies to solve the online scheduling problem: fast greedy algorithms based on the list scheduling concept or computing an optimal solution for the scheduling problem. The greedy approach was implemented using a modified version of the well-known HEFT list scheduler. Three extensions were developed to handle advance reservations and co-allocations. The greedy approach was compared to the computation of the optimal solution for the scheduling problem using genetic algorithms. However, experiments showed that the greedy algorithms provide in much shorter time comparable results. Any fast online scheduler may lead to resource fragmentation and, therefore, performance degradation. In order to deal with this problem, the online scheduling was enhanced to increase the utilization of the Grid using a background task to 11

12 1 Introduction optimize the schedule. To perform this optimization, a measure to rate the schedule is needed. When a request is rejected while in sum sufficient capacity was available, this rejection is caused by fragmentation. In such a case, the schedule can be called fragmented. However, a general quantification of the fragmentation of schedules was still missing. A new fragmentation measure for two-dimensional schedules was developed. Experiments showed the correlation between the rating of a schedule and the future rate of rejected jobs. The new measure was used to detect the degradation of the schedule. Furthermore, the detected extend of the fragmentation was then used as input parameter to guide the offline optimization. Three options to perform the offline scheduling are discussed in Chapter 5. The optimization can either be done for the schedule of a single resource only, assuming the Grid jobs to be fixed, or by negotiating the flexibility of the Grid jobs with the Grid broker. Another option is to make the Grid broker responsible for the optimization. Then, the knowledge about the Grid workflow structure and the Grid wide view allows better rearrangements, but again the view on the state of the local schedules is hidden. For all three options, the needed changes in the admission protocol and algorithms to efficiently reoptimize the schedule are presented and evaluated by simulations. The fourth part of the thesis deals with the important complex of rescue scheduling. If an SLA was given in advance, the Grid broker also has to cope with unexpected resource downtimes. Most failure handling mechanisms in the high performance computing domain cover mainly the currently running jobs. However, in an advance reservation environment, failure handling has to deal with the planned, but not yet started jobs, too. This failure handling requires finding an alternative schedule for the accepted requests. The general idea is to remap the jobs that are likely to be affected by the failure. As the failure is only temporary, remapping too many jobs in advance also leads to performance losses. The new concept of a remapping horizon is introduced which will be repeatedly reapplied during the downtime to determine the jobs to be remapped. Two techniques to determine the remapping horizon are developed and compared by simulations in Chapter 6. All algorithms developed in this thesis are evaluated using simulations since there is only little information on real world advance reservation enabled systems and Grid workflow systems available. Hence, a synthetic workload generator had to be developed. Furthermore, except for the VRM framework presented in this thesis, no installation in productive use is available combining both concepts. The newly developed workload generator combines the techniques used to generate task graph to benchmark scheduling in parallel computer, well known job size and arrival distributions from queuing systems, and new models for advance reservation. The remainder of this thesis is organized as follows: In the next chapter, the Grid environment and model assumptions used her are presented together with the strategies to manage advance reservations and the new fragmentation measure. 12

13 The following chapter covers one stage in the lifecycle each: Specification, online scheduling, offline scheduling, and rescue scheduling. In each of these chapters the specific related work is presented together with newly developed algorithms and their evaluation. 13

14

15 2 Coupling Resources in the Grid fast specifies admission input optimize schedule handle resource downtime The basic idea of the Grid is to unify the access to distributed high-performance compute resources. As the resources are not only distributed, but also belong to multiple, even competing organizations, new means of management are needed. In this chapter, the general requirements to form a Grid of a bunch of distributed computers are described. Additionally, the specific properties of the Grid model and the model assumptions used in this thesis are defined. As described before, one of the key model assumptions is the use of advance reservation in the Grid. This chapter introduces the basic concept, new data structures to manage advance reservations, and a new fragmentation measure to compare advance reservation schedules. 2.1 How to build up a Grid? The most cited definition of a computational Grid is the following by Foster [Fos02]: A Grid coordinates resources that are not subject to centralized control using standard, open, general-purpose protocols and interfaces to deliver nontrivial qualities of service. So one of the main benefits is to unify and simplify the user access to a lot of resources and add additional services. Nowadays, a typical user of scientific computing has access to more than one compute resource within his organization or by means of cooperation. But all resources provide their very specific user interface for submitting and monitoring jobs. Additionally, they may differ in organizational aspects like different prioritization strategies or accounting for resource usage. Especially, selecting the resource with the expected earliest finishing time can be hard in such a scenario. Using a Grid to couple all the resources, reduces the number of interfaces to only one and the Grid broker may also employ some user defined strategies (like earliest finish time or lowest cost) when seeking and booking a resource. While already coupling the resources in the same organization reduce the complexity of the allocation, the Grid can also span multiple organizations. By coupling two Grid domains, there may be only one cooperation contract necessary and all users of both sides participate through their usual interface. Grid technology is not only useful to manage compute resources; it can also be used to manage the distributed access to networks, storage systems, scientific equipment, and even meetings rooms or 3D visualization systems. In this thesis, 15

16 2 Coupling Resources in the Grid any type of resource together with the network bandwidth to connect them is considered. Beside the unified access, the added service level was another key element of the Grid definition. Grid broker may support service level agreements (SLA), i.e., they guarantee some properties for the execution of the job. SLAs can cover the state of the executing resources, e.g., security and availability. A widely used SLA aspect is to guarantee the execution of a job within a negotiated time frame and, especially, by some deadline. It has been shown that advance reservation is a key technology to enable SLAs in the Grid [SWY04, BHKL04]. In Section 2.2 the concept of advance reservation is explained in detail. Another added service is to provide virtual resources by combining the participating resources. This can be done by splitting up a parallel program and executing a part of it on each site at the very same time. For widely distributed Grids and tightly coupled parallel applications this slows the execution too much down, but for some setups this is faster than using fewer resources on a single site. As most calculations are part of a scientific workflow Grid broker may support the allocation of complex jobs with interdependent sub-jobs. These so-called Grid workflows define which output from previous sub-jobs is needed for each sub-job. The Grid broker then seeks an allocation for the required resources, providing the order and also allocates network bandwidth to transfer the data, if two sub-jobs are not on the same site. The sub-jobs of a Grid workflow may require different kind of resources, e.g., an astronomical workflow may use a telescope to gather input data, a high performance computer to process the telescope data and a 3D visualization system to present the result to the end user (see Figure 2.11 on page 35). Grid workflows may also contain sub-jobs to be executed at the very same time, so-called, co-allocations. Up to now, only benefits for the end user and the Grid broker are shown. Resource operator also benefit from participating in the Grid. Instead of negotiating and billing the access with every user individually, the resource operator just deals with one or more Grid broker representing a number of end user. Resource management systems for single sites usually use all available information like load on each node and whether a job can be postponed without violating a SLA. In the distributed setting, where the Grid broker as a high level resource management system is operated by another organization, the resource owner may restrict the access to this information. Especially, if he has contracts with other Grid brokers or local users, he doesn t want to give information about their jobs. Even the current load level of the resource can be used to reconstruct such information. A general goal of the resource operator participating in the Grid is to hide as much information as possible and to provide as little information as it is needed for the allocation process in order to still get jobs. The negotiation of SLAs also implies the negotiation of a payment for the service 16

17 2.2 Timing: Advance Reservation and a fine to be paid if the service was not fulfilled. In order to analyze the behavior of such a system a cost model is needed. The cost model defines how the user decisions are affected by different prices and how resource operators calculate offers in order to gain the maximum profit. Also the fine negotiation itself is a very complex topic. The fine should be very high to prevent the resource operator from deliberately violating the SLA and it should be reasonable low so that the resource operator does not face bankruptcy after each small failure. As these costs models are a complex topic on themselves, they are excluded in this thesis. A simple SLA model without quantifying the costs will be used instead. In this model, the users accept every solution which fulfills the specified requirements and constraints and the resource operators maximize their profit by solely increasing the utilization of their systems and the negotiated fine is assumed so high, that the resource operators will avoid a violation by all means. 2.2 Timing: Advance Reservation Most problems of sharing scarce resources are solved by queuing up the requests. The concept is the same for shopping in a grocery store as for high performance computing. The concept is very simple to realize and, by using a first-come-firstserve-strategy, also provides means to predict the actual start time. For combined transactions so called co-allocations with multiple providers, this concept is already too weak. If one wants to travel, it wouldn t be sufficient to queue up at the airline counter and then after arriving at the destination to enter the queue for the hotel without even knowing if there will be a room available on the same day. If one can choose between service providers, each using an independent queue, it is also hard to select the one which will be available first. This is also a commonly known problem in grocery stores. But in that setup, it is at least possible by earlier experiences and the transparent view on the shopping cart of the other customers to make a prediction of the waiting time. In general, the queues are not transparent and if there is no additional hint by the other users or the service provider, the wait time cannot be predicted. Hence, a user cannot compare independent resource providers. Another concept also used in the everyday life to avoid the problems of the queuing approach, is to negotiate the actual start time with the service provider. This means that the user contacts the service provider in advance, usually when the need for the service is in sight, and negotiates a start time in the future. Both parties agree that the service will be fulfilled then, i.e., the service provider will have all needed resources available and the user will show up to use the service. This advance reservation approach has a number of benefits for the user and 17

18 2 Coupling Resources in the Grid the service provider. By negotiating with different service providers, the user can coordinate the execution of co-allocations. In the travel example the usual approach is to book the flight and the hotel room in advance and if the rooms are only available on other dates, to adapt the flight date and vice versa. Another benefit for the user is that he can negotiate with multiple service providers for the same service and can compare the waiting time and other properties of the offer. The service provider increases the service level with advance reservation, as he can now guarantee that the service will be finished by some given deadline [WSV + 06]. If the book-ahead time the time between the negotiation and the execution of the service is long enough, the service provider can also adjust the provided resource capacity to meet only the requested capacity, e.g., the amount of raw material stocked or the number of employees working at a given time. On the other hand, the service provider can use the negotiation phase for load balancing, e.g., promoting phases with lower utilization. Using advance reservation also implies some drawbacks. The most obvious one is that it is much more complex to handle then the queuing approach. In the real world a queue is easily formed by the waiting customer and the service provider only has to provide some space for queuing. Even in a compute system, a queue of waiting jobs is just a simple data structure. For advance reservation, someone is needed to negotiate with the user together with some infrastructure to keep track of all the advance reservations and to identify free capacities in the future. Therefore, there is a significant overhead for the resource provider. In Section 2.2.2, data structures to handle advance reservations are discussed in detail. For planning in advance, the execution times of the services have to be known. Therefore, either the user or the service provider has to indicate the duration of the service usage. In the real world mostly the service providers know, how long the different kinds of services will take, in the world of high performance computing the user usually provides a prediction [FRS + 97, TEF05]. But there are also approaches to predict the execution time based on previous executions of the same parallel program [FRS + 97, Ang98, SFT98]. The prediction of the duration is usually inaccurate. If it is too short, the timetable with the negotiated reservations may break. This leads to a queue of waiting user or the violation of the negotiated SLAs. The service provider can also decide to stop the service after the negotiated duration, even if the service has not yet finished. Then only a single SLA is broken and all other reservations can be handled as planned. On the other hand, if the service usage finished earlier than negotiated, the booked resources are unused until the start time of the next reservation. The negotiation of the start time may lead to time frames without resource usage, which are too small to fulfill further incoming requests. Therefore, there is not only a fragmentation in the resource dimension, but also in the time dimension. In Section this two-dimensional fragmentation problem is discussed in depth. 18

19 2.2 Timing: Advance Reservation t book-start (j) d j t book-end (j) resource cj t start (j) t arr (j) time Figure 2.1: Advance reservation allows planning the execution of jobs in the future. In a system mixing queued jobs with advance reservations, the utilization drops due to the fixed advance reservation. The reservations obviously disturb the queue reordering (backfilling), often used in queuing based systems to increase the utilization [SFT00]. Sulistio and Buyya measured a utilization drop of 60-80% and increased wait times for the jobs in the queue. However, Heine et al. showed that by using a native planning based resource management system instead of an enhanced queuing based, the impact of the advance reservations is less dramatic [HHKS05]. In [Mac03] a comprehensive overview on resource management systems supporting advance reservation is given. Summarizing, advance reservation is an elaborated technology allowing the coordinated allocation of jobs. However, a lot of additional information is required and a processing overhead is introduced. Nonetheless, it is an important base technology to process complex Grid workflows and to guarantee SLAs in the Grid Formalization After introducing the basic concept of advance reservation, the following section provides a more formal view on the concept of advance reservation. A job j is any request for resource usage. It is specified by the resource type it requests T j, how much of this resources capacity will be used c j, and the duration of 19

20 2 Coupling Resources in the Grid the resource usage d j. The unit of c j depends on the type of the requested resource, e.g., number of CPUs for parallel computer, kbit/s bandwidth for networks, or just 0 or 1 for single-unit resources like 3D visualization systems. The duration is finite and the job will be executed once. Periodic executions are not considered in the resource manager, while the user may periodically submit the same job. Each resource r R is defined by its type T r and the available capacity c r. Network resources enable the communication between the resources of the other types. In a very abstract view, a network connects many different resources and in a very detailed view, a network consists of a number of links connecting a resource with a router or the routers within the network. Both views are not practical from the Grid broker s point of view. The abstract view is too generic and the actual mapping of a flow to the specific links should be left to the network resource management system. Thus, the Grid broker sees a network as a bunch of resources each providing the communication between a pair of other resources. The relations source, target : R N R provide the information which resources are connected by this network resource. All these resources will be managed by a single network RMS and reservations may interfere with each other even if they connect different endpoints. How advance reservation in a network can be done is discussed in detail in [Bur04] and [FFR + 04]. Independent of the type, jobs arrive at the resource management system (RMS) before their respective execution starts. This moment is called arrival time of the job t arr (j) (see Figure 2.1). The requester may also restrict the possible execution time by providing a booking interval with the start t book start (j) and end time t book end (j). In order to get at least one possible start time, t book start (j) + d j t book end has to hold. If no booking interval is specified t book start (j) = and t book end (j) = will be used. The RMS will then determine a matching start time t start (j) for the job, such that t book start (j) t start (j) and t start (j) + d j t book end. The difference between the arrival time and the actual start time is called bookahead time t book ahead (j) = t start (j) t arr (j). To reduce the complexity of the scheduling process the RMS may impose a maximum book ahead time ˆt book ahead. The RMS has to keep track of all already reserved jobs J r, the respective start times, and resource allocations. In general, only as much capacity can be reserved as is available on the resource: t : c r j J r t start(j) t t start(j)+d j c j Hence, the scheduling problem can be formulated as: Determine a t start (j) such 20

Contributions to Gang Scheduling

Contributions to Gang Scheduling CHAPTER 7 Contributions to Gang Scheduling In this Chapter, we present two techniques to improve Gang Scheduling policies by adopting the ideas of this Thesis. The first one, Performance- Driven Gang Scheduling,

More information

159.735. Final Report. Cluster Scheduling. Submitted by: Priti Lohani 04244354

159.735. Final Report. Cluster Scheduling. Submitted by: Priti Lohani 04244354 159.735 Final Report Cluster Scheduling Submitted by: Priti Lohani 04244354 1 Table of contents: 159.735... 1 Final Report... 1 Cluster Scheduling... 1 Table of contents:... 2 1. Introduction:... 3 1.1

More information

Resource Allocation Schemes for Gang Scheduling

Resource Allocation Schemes for Gang Scheduling Resource Allocation Schemes for Gang Scheduling B. B. Zhou School of Computing and Mathematics Deakin University Geelong, VIC 327, Australia D. Walsh R. P. Brent Department of Computer Science Australian

More information

Scheduling. from CPUs to Clusters to Grids. MCSN N. Tonellotto Complements of Distributed Enabling Platforms

Scheduling. from CPUs to Clusters to Grids. MCSN N. Tonellotto Complements of Distributed Enabling Platforms Scheduling from CPUs to Clusters to Grids 1 Outline Terminology CPU Scheduling Real-time Scheduling Cluster Scheduling Grid Scheduling Cloud Scheduling 2 General Scheduling refers to allocate limited resources

More information

Evaluation of Selection Methods for Global Mobility Management Software

Evaluation of Selection Methods for Global Mobility Management Software Leibniz Universität Hannover Wirtschaftswissenschaftliche Fakultät Institut für Wirtschaftsinformatik Evaluation of Selection Methods for Global Mobility Management Software Masterarbeit zur Erlangung

More information

Power Management in Cloud Computing using Green Algorithm. -Kushal Mehta COP 6087 University of Central Florida

Power Management in Cloud Computing using Green Algorithm. -Kushal Mehta COP 6087 University of Central Florida Power Management in Cloud Computing using Green Algorithm -Kushal Mehta COP 6087 University of Central Florida Motivation Global warming is the greatest environmental challenge today which is caused by

More information

Introduction to LAN/WAN. Network Layer

Introduction to LAN/WAN. Network Layer Introduction to LAN/WAN Network Layer Topics Introduction (5-5.1) Routing (5.2) (The core) Internetworking (5.5) Congestion Control (5.3) Network Layer Design Isues Store-and-Forward Packet Switching Services

More information

Fair Scheduling Algorithm with Dynamic Load Balancing Using In Grid Computing

Fair Scheduling Algorithm with Dynamic Load Balancing Using In Grid Computing Research Inventy: International Journal Of Engineering And Science Vol.2, Issue 10 (April 2013), Pp 53-57 Issn(e): 2278-4721, Issn(p):2319-6483, Www.Researchinventy.Com Fair Scheduling Algorithm with Dynamic

More information

Overlapping Data Transfer With Application Execution on Clusters

Overlapping Data Transfer With Application Execution on Clusters Overlapping Data Transfer With Application Execution on Clusters Karen L. Reid and Michael Stumm reid@cs.toronto.edu stumm@eecg.toronto.edu Department of Computer Science Department of Electrical and Computer

More information

An Enterprise Modeling Framework for Banks using. Algebraic Graph Transformation

An Enterprise Modeling Framework for Banks using. Algebraic Graph Transformation An Enterprise Modeling Framework for Banks using Algebraic Graph Transformation vorgelegt von Diplom-Wirtschaftsinformatiker Christoph Brandt aus Berlin-Lichterfelde von der Fakultät IV - Elektrotechnik

More information

CHAPTER 1 INTRODUCTION

CHAPTER 1 INTRODUCTION CHAPTER 1 INTRODUCTION 1.1 Background The command over cloud computing infrastructure is increasing with the growing demands of IT infrastructure during the changed business scenario of the 21 st Century.

More information

Grid Computing Approach for Dynamic Load Balancing

Grid Computing Approach for Dynamic Load Balancing International Journal of Computer Sciences and Engineering Open Access Review Paper Volume-4, Issue-1 E-ISSN: 2347-2693 Grid Computing Approach for Dynamic Load Balancing Kapil B. Morey 1*, Sachin B. Jadhav

More information

SLA-based Admission Control for a Software-as-a-Service Provider in Cloud Computing Environments

SLA-based Admission Control for a Software-as-a-Service Provider in Cloud Computing Environments SLA-based Admission Control for a Software-as-a-Service Provider in Cloud Computing Environments Linlin Wu, Saurabh Kumar Garg, and Rajkumar Buyya Cloud Computing and Distributed Systems (CLOUDS) Laboratory

More information

Software Engineering Prof. N.L. Sarda Computer Science & Engineering Indian Institute of Technology, Bombay Lecture-4 Overview of Phases (Part - II)

Software Engineering Prof. N.L. Sarda Computer Science & Engineering Indian Institute of Technology, Bombay Lecture-4 Overview of Phases (Part - II) Software Engineering Prof. N.L. Sarda Computer Science & Engineering Indian Institute of Technology, Bombay Lecture-4 Overview of Phases (Part - II) We studied the problem definition phase, with which

More information

Visualization methods for patent data

Visualization methods for patent data Visualization methods for patent data Treparel 2013 Dr. Anton Heijs (CTO & Founder) Delft, The Netherlands Introduction Treparel can provide advanced visualizations for patent data. This document describes

More information

Advance Reservations of Bandwidth in Computer Networks

Advance Reservations of Bandwidth in Computer Networks Advance Reservations of Bandwidth in Computer Networks vorgelegt von Diplom-Informatiker Lars-Olof Burchard aus Berlin von der Fakultät IV - Elektrotechnik und Informatik der Technischen Universität Berlin

More information

GEDAE TM - A Graphical Programming and Autocode Generation Tool for Signal Processor Applications

GEDAE TM - A Graphical Programming and Autocode Generation Tool for Signal Processor Applications GEDAE TM - A Graphical Programming and Autocode Generation Tool for Signal Processor Applications Harris Z. Zebrowitz Lockheed Martin Advanced Technology Laboratories 1 Federal Street Camden, NJ 08102

More information

4.2: Multimedia File Systems Traditional File Systems. Multimedia File Systems. Multimedia File Systems. Disk Scheduling

4.2: Multimedia File Systems Traditional File Systems. Multimedia File Systems. Multimedia File Systems. Disk Scheduling Chapter 2: Representation of Multimedia Data Chapter 3: Multimedia Systems Communication Aspects and Services Chapter 4: Multimedia Systems Storage Aspects Optical Storage Media Multimedia File Systems

More information

Scheduling in a Virtual Enterprise in the Service Sector

Scheduling in a Virtual Enterprise in the Service Sector Scheduling in a Virtual Enterprise in the Service Sector Florian Kandler Electronic Commerce Competence Center, Donau-City-Strasse, 7 Vienna, Austria florian.kandler@ec,at http://www.ec.at/ Abstract. The

More information

Communication Protocol

Communication Protocol Analysis of the NXT Bluetooth Communication Protocol By Sivan Toledo September 2006 The NXT supports Bluetooth communication between a program running on the NXT and a program running on some other Bluetooth

More information

11.1 inspectit. 11.1. inspectit

11.1 inspectit. 11.1. inspectit 11.1. inspectit Figure 11.1. Overview on the inspectit components [Siegl and Bouillet 2011] 11.1 inspectit The inspectit monitoring tool (website: http://www.inspectit.eu/) has been developed by NovaTec.

More information

Windows Server Performance Monitoring

Windows Server Performance Monitoring Spot server problems before they are noticed The system s really slow today! How often have you heard that? Finding the solution isn t so easy. The obvious questions to ask are why is it running slowly

More information

Analysis of Micromouse Maze Solving Algorithms

Analysis of Micromouse Maze Solving Algorithms 1 Analysis of Micromouse Maze Solving Algorithms David M. Willardson ECE 557: Learning from Data, Spring 2001 Abstract This project involves a simulation of a mouse that is to find its way through a maze.

More information

The Importance of Software License Server Monitoring

The Importance of Software License Server Monitoring The Importance of Software License Server Monitoring NetworkComputer Meeting The Job Scheduling Challenges of Organizations of All Sizes White Paper Introduction Every semiconductor design group uses a

More information

Load Balancing on a Non-dedicated Heterogeneous Network of Workstations

Load Balancing on a Non-dedicated Heterogeneous Network of Workstations Load Balancing on a Non-dedicated Heterogeneous Network of Workstations Dr. Maurice Eggen Nathan Franklin Department of Computer Science Trinity University San Antonio, Texas 78212 Dr. Roger Eggen Department

More information

WHITE PAPER. iet ITSM Enables Enhanced Service Management

WHITE PAPER. iet ITSM Enables Enhanced Service Management iet ITSM Enables Enhanced Service Management iet ITSM Enables Enhanced Service Management Need for IT Service Management The focus within the vast majority of large and medium-size companies has shifted

More information

Agile support with Kanban some tips and tricks By Tomas Björkholm

Agile support with Kanban some tips and tricks By Tomas Björkholm Agile support with Kanban some tips and tricks By Tomas Björkholm Foreword A year ago I held an Open Space at Scrum Gathering in Stockholm about Agile Support. I have since received several requests to

More information

CHAPTER 2. QoS ROUTING AND ITS ROLE IN QOS PARADIGM

CHAPTER 2. QoS ROUTING AND ITS ROLE IN QOS PARADIGM CHAPTER 2 QoS ROUTING AND ITS ROLE IN QOS PARADIGM 22 QoS ROUTING AND ITS ROLE IN QOS PARADIGM 2.1 INTRODUCTION As the main emphasis of the present research work is on achieving QoS in routing, hence this

More information

1. The subnet must prevent additional packets from entering the congested region until those already present can be processed.

1. The subnet must prevent additional packets from entering the congested region until those already present can be processed. Congestion Control When one part of the subnet (e.g. one or more routers in an area) becomes overloaded, congestion results. Because routers are receiving packets faster than they can forward them, one

More information

International Journal of Advance Research in Computer Science and Management Studies

International Journal of Advance Research in Computer Science and Management Studies Volume 3, Issue 6, June 2015 ISSN: 2321 7782 (Online) International Journal of Advance Research in Computer Science and Management Studies Research Article / Survey Paper / Case Study Available online

More information

Questions? Assignment. Techniques for Gathering Requirements. Gathering and Analysing Requirements

Questions? Assignment. Techniques for Gathering Requirements. Gathering and Analysing Requirements Questions? Assignment Why is proper project management important? What is goal of domain analysis? What is the difference between functional and non- functional requirements? Why is it important for requirements

More information

A Tool for Generating Partition Schedules of Multiprocessor Systems

A Tool for Generating Partition Schedules of Multiprocessor Systems A Tool for Generating Partition Schedules of Multiprocessor Systems Hans-Joachim Goltz and Norbert Pieth Fraunhofer FIRST, Berlin, Germany {hans-joachim.goltz,nobert.pieth}@first.fraunhofer.de Abstract.

More information

Grid Scheduling Dictionary of Terms and Keywords

Grid Scheduling Dictionary of Terms and Keywords Grid Scheduling Dictionary Working Group M. Roehrig, Sandia National Laboratories W. Ziegler, Fraunhofer-Institute for Algorithms and Scientific Computing Document: Category: Informational June 2002 Status

More information

The Importance of Software License Server Monitoring

The Importance of Software License Server Monitoring The Importance of Software License Server Monitoring NetworkComputer How Shorter Running Jobs Can Help In Optimizing Your Resource Utilization White Paper Introduction Semiconductor companies typically

More information

Requirements engineering

Requirements engineering Learning Unit 2 Requirements engineering Contents Introduction............................................... 21 2.1 Important concepts........................................ 21 2.1.1 Stakeholders and

More information

Service Level Agreements based on Business Process Modeling

Service Level Agreements based on Business Process Modeling Service Level Agreements based on Business Process Modeling Holger Schmidt Munich Network Management Team University of Munich, Dept. of CS Oettingenstr. 67, 80538 Munich, Germany Email: schmidt@informatik.uni-muenchen.de

More information

Heterogeneous Workload Consolidation for Efficient Management of Data Centers in Cloud Computing

Heterogeneous Workload Consolidation for Efficient Management of Data Centers in Cloud Computing Heterogeneous Workload Consolidation for Efficient Management of Data Centers in Cloud Computing Deep Mann ME (Software Engineering) Computer Science and Engineering Department Thapar University Patiala-147004

More information

Optimizing Shared Resource Contention in HPC Clusters

Optimizing Shared Resource Contention in HPC Clusters Optimizing Shared Resource Contention in HPC Clusters Sergey Blagodurov Simon Fraser University Alexandra Fedorova Simon Fraser University Abstract Contention for shared resources in HPC clusters occurs

More information

Managing Cloud Service Provisioning and SLA Enforcement via Holistic Monitoring Techniques Vincent C. Emeakaroha

Managing Cloud Service Provisioning and SLA Enforcement via Holistic Monitoring Techniques Vincent C. Emeakaroha Managing Cloud Service Provisioning and SLA Enforcement via Holistic Monitoring Techniques Vincent C. Emeakaroha Matrikelnr: 0027525 vincent@infosys.tuwien.ac.at Supervisor: Univ.-Prof. Dr. Schahram Dustdar

More information

Linux Driver Devices. Why, When, Which, How?

Linux Driver Devices. Why, When, Which, How? Bertrand Mermet Sylvain Ract Linux Driver Devices. Why, When, Which, How? Since its creation in the early 1990 s Linux has been installed on millions of computers or embedded systems. These systems may

More information

Discrete-Event Simulation

Discrete-Event Simulation Discrete-Event Simulation Prateek Sharma Abstract: Simulation can be regarded as the emulation of the behavior of a real-world system over an interval of time. The process of simulation relies upon the

More information

A Comparative Performance Analysis of Load Balancing Algorithms in Distributed System using Qualitative Parameters

A Comparative Performance Analysis of Load Balancing Algorithms in Distributed System using Qualitative Parameters A Comparative Performance Analysis of Load Balancing Algorithms in Distributed System using Qualitative Parameters Abhijit A. Rajguru, S.S. Apte Abstract - A distributed system can be viewed as a collection

More information

Efficient DNS based Load Balancing for Bursty Web Application Traffic

Efficient DNS based Load Balancing for Bursty Web Application Traffic ISSN Volume 1, No.1, September October 2012 International Journal of Science the and Internet. Applied However, Information this trend leads Technology to sudden burst of Available Online at http://warse.org/pdfs/ijmcis01112012.pdf

More information

Satellite-UMTS - Specification of Protocols and Traffic Performance Analysis

Satellite-UMTS - Specification of Protocols and Traffic Performance Analysis Satellite-UMTS - Specification of Protocols and Traffic Performance Analysis Von der Fakultat fur Elektrotechnik und Informationstechnik der Rheinisch-Westfalichen Technischen Hochschule Aachen zur Erlangung

More information

NOVEL PRIORITISED EGPRS MEDIUM ACCESS REGIME FOR REDUCED FILE TRANSFER DELAY DURING CONGESTED PERIODS

NOVEL PRIORITISED EGPRS MEDIUM ACCESS REGIME FOR REDUCED FILE TRANSFER DELAY DURING CONGESTED PERIODS NOVEL PRIORITISED EGPRS MEDIUM ACCESS REGIME FOR REDUCED FILE TRANSFER DELAY DURING CONGESTED PERIODS D. Todinca, P. Perry and J. Murphy Dublin City University, Ireland ABSTRACT The goal of this paper

More information

Load balancing model for Cloud Data Center ABSTRACT:

Load balancing model for Cloud Data Center ABSTRACT: Load balancing model for Cloud Data Center ABSTRACT: Cloud data center management is a key problem due to the numerous and heterogeneous strategies that can be applied, ranging from the VM placement to

More information

MINIMIZING STORAGE COST IN CLOUD COMPUTING ENVIRONMENT

MINIMIZING STORAGE COST IN CLOUD COMPUTING ENVIRONMENT MINIMIZING STORAGE COST IN CLOUD COMPUTING ENVIRONMENT 1 SARIKA K B, 2 S SUBASREE 1 Department of Computer Science, Nehru College of Engineering and Research Centre, Thrissur, Kerala 2 Professor and Head,

More information

Binary search tree with SIMD bandwidth optimization using SSE

Binary search tree with SIMD bandwidth optimization using SSE Binary search tree with SIMD bandwidth optimization using SSE Bowen Zhang, Xinwei Li 1.ABSTRACT In-memory tree structured index search is a fundamental database operation. Modern processors provide tremendous

More information

An Analysis of the B2B E-Contracting Domain - Paradigms and Required Technology 1

An Analysis of the B2B E-Contracting Domain - Paradigms and Required Technology 1 An Analysis of the B2B E-Contracting Domain - Paradigms and Required Technology 1 Samuil Angelov and Paul Grefen Department of Technology Management, Eindhoven University of Technology, P.O. Box 513, 5600

More information

Model-Driven Scientific Workflow Engineering

Model-Driven Scientific Workflow Engineering A Framework for Model-Driven Scientific Workflow Engineering Dipl-Inform Guido Scherp Dissertation zur Erlangung des akademischen Grades Doktor der Ingenieurwissenschaften (Dr-Ing) der Technischen Fakultat

More information

BSC vision on Big Data and extreme scale computing

BSC vision on Big Data and extreme scale computing BSC vision on Big Data and extreme scale computing Jesus Labarta, Eduard Ayguade,, Fabrizio Gagliardi, Rosa M. Badia, Toni Cortes, Jordi Torres, Adrian Cristal, Osman Unsal, David Carrera, Yolanda Becerra,

More information

Achieving Nanosecond Latency Between Applications with IPC Shared Memory Messaging

Achieving Nanosecond Latency Between Applications with IPC Shared Memory Messaging Achieving Nanosecond Latency Between Applications with IPC Shared Memory Messaging In some markets and scenarios where competitive advantage is all about speed, speed is measured in micro- and even nano-seconds.

More information

Journal of Computer and System Sciences

Journal of Computer and System Sciences Journal of Computer and System Sciences 78 (2012) 1280 1299 Contents lists available at SciVerse ScienceDirect Journal of Computer and System Sciences www.elsevier.com/locate/jcss SLA-based admission control

More information

Guideline for stresstest Page 1 of 6. Stress test

Guideline for stresstest Page 1 of 6. Stress test Guideline for stresstest Page 1 of 6 Stress test Objective: Show unacceptable problems with high parallel load. Crash, wrong processing, slow processing. Test Procedure: Run test cases with maximum number

More information

7 Conclusions and suggestions for further research

7 Conclusions and suggestions for further research 7 Conclusions and suggestions for further research This research has devised an approach to analyzing system-level coordination from the point of view of product architecture. The analysis was conducted

More information

A Scalable Network Monitoring and Bandwidth Throttling System for Cloud Computing

A Scalable Network Monitoring and Bandwidth Throttling System for Cloud Computing A Scalable Network Monitoring and Bandwidth Throttling System for Cloud Computing N.F. Huysamen and A.E. Krzesinski Department of Mathematical Sciences University of Stellenbosch 7600 Stellenbosch, South

More information

Smart Queue Scheduling for QoS Spring 2001 Final Report

Smart Queue Scheduling for QoS Spring 2001 Final Report ENSC 833-3: NETWORK PROTOCOLS AND PERFORMANCE CMPT 885-3: SPECIAL TOPICS: HIGH-PERFORMANCE NETWORKS Smart Queue Scheduling for QoS Spring 2001 Final Report By Haijing Fang(hfanga@sfu.ca) & Liu Tang(llt@sfu.ca)

More information

Parallel Ray Tracing using MPI: A Dynamic Load-balancing Approach

Parallel Ray Tracing using MPI: A Dynamic Load-balancing Approach Parallel Ray Tracing using MPI: A Dynamic Load-balancing Approach S. M. Ashraful Kadir 1 and Tazrian Khan 2 1 Scientific Computing, Royal Institute of Technology (KTH), Stockholm, Sweden smakadir@csc.kth.se,

More information

WORKFLOW ENGINE FOR CLOUDS

WORKFLOW ENGINE FOR CLOUDS WORKFLOW ENGINE FOR CLOUDS By SURAJ PANDEY, DILEBAN KARUNAMOORTHY, and RAJKUMAR BUYYA Prepared by: Dr. Faramarz Safi Islamic Azad University, Najafabad Branch, Esfahan, Iran. Workflow Engine for clouds

More information

Big, Fast Routers. Router Architecture

Big, Fast Routers. Router Architecture Big, Fast Routers Dave Andersen CMU CS 15-744 Data Plane Router Architecture How packets get forwarded Control Plane How routing protocols establish routes/etc. 1 Processing: Fast Path vs. Slow Path Optimize

More information

TIME DEPENDENT PRIORITIES IN CALL CENTERS

TIME DEPENDENT PRIORITIES IN CALL CENTERS TIME DEPENDENT PRIORITIES IN CALL CENTERS LASSAAD ESSAFI and GUNTER BOLCH Institute of Computer Science University of Erlangen-Nuremberg Martensstrasse 3, D-91058 Erlangen, Germany E-mail : lassaad@essafi.de,

More information

A Multi Agent System for MS Windows using Matlab-enabled Agents

A Multi Agent System for MS Windows using Matlab-enabled Agents A Multi Agent System for MS Windows using Matlab-enabled Agents Th.M. Hupkens a R.Thaens b a Royal Netherlands Naval College, Combat Systems Department, P.O. Box 10000, 1780 CA Den Helder, The Netherlands

More information

Comparison of Request Admission Based Performance Isolation Approaches in Multi-tenant SaaS Applications

Comparison of Request Admission Based Performance Isolation Approaches in Multi-tenant SaaS Applications Comparison of Request Admission Based Performance Isolation Approaches in Multi-tenant SaaS Applications Rouven Kreb 1 and Manuel Loesch 2 1 SAP AG, Walldorf, Germany 2 FZI Research Center for Information

More information

Load Balancing. Load Balancing 1 / 24

Load Balancing. Load Balancing 1 / 24 Load Balancing Backtracking, branch & bound and alpha-beta pruning: how to assign work to idle processes without much communication? Additionally for alpha-beta pruning: implementing the young-brothers-wait

More information

Software Development for Multiple OEMs Using Tool Configured Middleware for CAN Communication

Software Development for Multiple OEMs Using Tool Configured Middleware for CAN Communication 01PC-422 Software Development for Multiple OEMs Using Tool Configured Middleware for CAN Communication Pascal Jost IAS, University of Stuttgart, Germany Stephan Hoffmann Vector CANtech Inc., USA Copyright

More information

Praktikum Wissenschaftliches Rechnen (Performance-optimized optimized Programming)

Praktikum Wissenschaftliches Rechnen (Performance-optimized optimized Programming) Praktikum Wissenschaftliches Rechnen (Performance-optimized optimized Programming) Dynamic Load Balancing Dr. Ralf-Peter Mundani Center for Simulation Technology in Engineering Technische Universität München

More information

Performance Monitoring of Parallel Scientific Applications

Performance Monitoring of Parallel Scientific Applications Performance Monitoring of Parallel Scientific Applications Abstract. David Skinner National Energy Research Scientific Computing Center Lawrence Berkeley National Laboratory This paper introduces an infrastructure

More information

EFFICIENT SCHEDULING STRATEGY USING COMMUNICATION AWARE SCHEDULING FOR PARALLEL JOBS IN CLUSTERS

EFFICIENT SCHEDULING STRATEGY USING COMMUNICATION AWARE SCHEDULING FOR PARALLEL JOBS IN CLUSTERS EFFICIENT SCHEDULING STRATEGY USING COMMUNICATION AWARE SCHEDULING FOR PARALLEL JOBS IN CLUSTERS A.Neela madheswari 1 and R.S.D.Wahida Banu 2 1 Department of Information Technology, KMEA Engineering College,

More information

Chapter 14: Distributed Operating Systems

Chapter 14: Distributed Operating Systems Chapter 14: Distributed Operating Systems Chapter 14: Distributed Operating Systems Motivation Types of Distributed Operating Systems Network Structure Network Topology Communication Structure Communication

More information

Self-Tuning Job Scheduling Strategies for the Resource Management of HPC Systems and Computational Grids

Self-Tuning Job Scheduling Strategies for the Resource Management of HPC Systems and Computational Grids Self-Tuning Job Scheduling Strategies for the Resource Management of HPC Systems and Computational Grids Dissertation von Achim Streit Schriftliche Arbeit zur Erlangung des Grades eines Doktors der Naturwissenschaften

More information

WHITE PAPER WORK PROCESS AND TECHNOLOGIES FOR MAGENTO PERFORMANCE (BASED ON FLIGHT CLUB) June, 2014. Project Background

WHITE PAPER WORK PROCESS AND TECHNOLOGIES FOR MAGENTO PERFORMANCE (BASED ON FLIGHT CLUB) June, 2014. Project Background WHITE PAPER WORK PROCESS AND TECHNOLOGIES FOR MAGENTO PERFORMANCE (BASED ON FLIGHT CLUB) June, 2014 Project Background Flight Club is the world s leading sneaker marketplace specialising in storing, shipping,

More information

An objective comparison test of workload management systems

An objective comparison test of workload management systems An objective comparison test of workload management systems Igor Sfiligoi 1 and Burt Holzman 1 1 Fermi National Accelerator Laboratory, Batavia, IL 60510, USA E-mail: sfiligoi@fnal.gov Abstract. The Grid

More information

Measuring the Performance of an Agent

Measuring the Performance of an Agent 25 Measuring the Performance of an Agent The rational agent that we are aiming at should be successful in the task it is performing To assess the success we need to have a performance measure What is rational

More information

Study of Various Load Balancing Techniques in Cloud Environment- A Review

Study of Various Load Balancing Techniques in Cloud Environment- A Review International Journal of Computer Sciences and Engineering Open Access Review Paper Volume-4, Issue-04 E-ISSN: 2347-2693 Study of Various Load Balancing Techniques in Cloud Environment- A Review Rajdeep

More information

Agenda. Distributed System Structures. Why Distributed Systems? Motivation

Agenda. Distributed System Structures. Why Distributed Systems? Motivation Agenda Distributed System Structures CSCI 444/544 Operating Systems Fall 2008 Motivation Network structure Fundamental network services Sockets and ports Client/server model Remote Procedure Call (RPC)

More information

Computer Architecture

Computer Architecture Computer Architecture Slide Sets WS 2013/2014 Prof. Dr. Uwe Brinkschulte M.Sc. Benjamin Betting Part 11 Memory Management Computer Architecture Part 11 page 1 of 44 Prof. Dr. Uwe Brinkschulte, M.Sc. Benjamin

More information

Future Generation Computer Systems. Deadline-constrained workflow scheduling algorithms for Infrastructure as a Service Clouds

Future Generation Computer Systems. Deadline-constrained workflow scheduling algorithms for Infrastructure as a Service Clouds Future Generation Computer Systems 29 (2013) 158 169 Contents lists available at SciVerse ScienceDirect Future Generation Computer Systems journal homepage: www.elsevier.com/locate/fgcs Deadline-constrained

More information

Ten Steps to Comprehensive Project Portfolio Management Part 3 Projects, Programs, Portfolios and Strategic Direction By R.

Ten Steps to Comprehensive Project Portfolio Management Part 3 Projects, Programs, Portfolios and Strategic Direction By R. August 2007 Ten Steps to Comprehensive Project Portfolio Management Part 3 Projects, Programs, Portfolios and Strategic Direction By R. Max Wideman This series of papers has been developed from our work

More information

Real-Time Scheduling 1 / 39

Real-Time Scheduling 1 / 39 Real-Time Scheduling 1 / 39 Multiple Real-Time Processes A runs every 30 msec; each time it needs 10 msec of CPU time B runs 25 times/sec for 15 msec C runs 20 times/sec for 5 msec For our equation, A

More information

3D Interactive Information Visualization: Guidelines from experience and analysis of applications

3D Interactive Information Visualization: Guidelines from experience and analysis of applications 3D Interactive Information Visualization: Guidelines from experience and analysis of applications Richard Brath Visible Decisions Inc., 200 Front St. W. #2203, Toronto, Canada, rbrath@vdi.com 1. EXPERT

More information

Chapter 18: Database System Architectures. Centralized Systems

Chapter 18: Database System Architectures. Centralized Systems Chapter 18: Database System Architectures! Centralized Systems! Client--Server Systems! Parallel Systems! Distributed Systems! Network Types 18.1 Centralized Systems! Run on a single computer system and

More information

The Trip Scheduling Problem

The Trip Scheduling Problem The Trip Scheduling Problem Claudia Archetti Department of Quantitative Methods, University of Brescia Contrada Santa Chiara 50, 25122 Brescia, Italy Martin Savelsbergh School of Industrial and Systems

More information

Quality of Service Guarantees for Cloud Services

Quality of Service Guarantees for Cloud Services Quality of Service Guarantees for Cloud Services CS848 Project presentation by Alexey Karyakin David R. Cheriton School of Computer Science University of Waterloo March 2010 Outline 1. Performance of cloud

More information

Optimized Scheduling in Real-Time Environments with Column Generation

Optimized Scheduling in Real-Time Environments with Column Generation JG U JOHANNES GUTENBERG UNIVERSITAT 1^2 Optimized Scheduling in Real-Time Environments with Column Generation Dissertation zur Erlangung des Grades,.Doktor der Naturwissenschaften" am Fachbereich Physik,

More information

RESOURCE CO-ALLOCATION ALGORITHMS IN DISTRIBUTED JOB BATCH SCHEDULING Victor V. Toporkov, Alexander Bobchenkov, Dmitry Yemelyanov and Anna Toporkova

RESOURCE CO-ALLOCATION ALGORITHMS IN DISTRIBUTED JOB BATCH SCHEDULING Victor V. Toporkov, Alexander Bobchenkov, Dmitry Yemelyanov and Anna Toporkova RESOURCE CO-ALLOCATION ALGORITHMS IN DISTRIBUTED JOB BATCH SCHEDULING Victor V. Toporkov, Alexander Bobchenkov, Dmitry Yemelyanov and Anna Toporkova Summary In this work, we present slot selection algorithms

More information

REVIEW PAPER ON PERFORMANCE OF RESTFUL WEB SERVICES

REVIEW PAPER ON PERFORMANCE OF RESTFUL WEB SERVICES REVIEW PAPER ON PERFORMANCE OF RESTFUL WEB SERVICES Miss.Monali K.Narse 1,Chaitali S.Suratkar 2, Isha M.Shirbhate 3 1 B.E, I.T, JDIET, Yavatmal, Maharashtra, India, monalinarse9990@gmail.com 2 Assistant

More information

Prescriptive Analytics. A business guide

Prescriptive Analytics. A business guide Prescriptive Analytics A business guide May 2014 Contents 3 The Business Value of Prescriptive Analytics 4 What is Prescriptive Analytics? 6 Prescriptive Analytics Methods 7 Integration 8 Business Applications

More information

Chapter 5 Linux Load Balancing Mechanisms

Chapter 5 Linux Load Balancing Mechanisms Chapter 5 Linux Load Balancing Mechanisms Load balancing mechanisms in multiprocessor systems have two compatible objectives. One is to prevent processors from being idle while others processors still

More information

A COOL AND PRACTICAL ALTERNATIVE TO TRADITIONAL HASH TABLES

A COOL AND PRACTICAL ALTERNATIVE TO TRADITIONAL HASH TABLES A COOL AND PRACTICAL ALTERNATIVE TO TRADITIONAL HASH TABLES ULFAR ERLINGSSON, MARK MANASSE, FRANK MCSHERRY MICROSOFT RESEARCH SILICON VALLEY MOUNTAIN VIEW, CALIFORNIA, USA ABSTRACT Recent advances in the

More information

Technical Investigation of Computational Resource Interdependencies

Technical Investigation of Computational Resource Interdependencies Technical Investigation of Computational Resource Interdependencies By Lars-Eric Windhab Table of Contents 1. Introduction and Motivation... 2 2. Problem to be solved... 2 3. Discussion of design choices...

More information

walberla: Towards an Adaptive, Dynamically Load-Balanced, Massively Parallel Lattice Boltzmann Fluid Simulation

walberla: Towards an Adaptive, Dynamically Load-Balanced, Massively Parallel Lattice Boltzmann Fluid Simulation walberla: Towards an Adaptive, Dynamically Load-Balanced, Massively Parallel Lattice Boltzmann Fluid Simulation SIAM Parallel Processing for Scientific Computing 2012 February 16, 2012 Florian Schornbaum,

More information

A SIMULATION STUDY FOR DYNAMIC FLEXIBLE JOB SHOP SCHEDULING WITH SEQUENCE-DEPENDENT SETUP TIMES

A SIMULATION STUDY FOR DYNAMIC FLEXIBLE JOB SHOP SCHEDULING WITH SEQUENCE-DEPENDENT SETUP TIMES A SIMULATION STUDY FOR DYNAMIC FLEXIBLE JOB SHOP SCHEDULING WITH SEQUENCE-DEPENDENT SETUP TIMES by Zakaria Yahia Abdelrasol Abdelgawad A Thesis Submitted to the Faculty of Engineering at Cairo University

More information

(Refer Slide Time 00:56)

(Refer Slide Time 00:56) Software Engineering Prof.N. L. Sarda Computer Science & Engineering Indian Institute of Technology, Bombay Lecture-12 Data Modelling- ER diagrams, Mapping to relational model (Part -II) We will continue

More information

THE BENEFITS OF SIGNAL GROUP ORIENTED CONTROL

THE BENEFITS OF SIGNAL GROUP ORIENTED CONTROL THE BENEFITS OF SIGNAL GROUP ORIENTED CONTROL Authors: Robbin Blokpoel 1 and Siebe Turksma 2 1: Traffic Engineering Researcher at Peek Traffic, robbin.blokpoel@peektraffic.nl 2: Product manager research

More information

An approach to grid scheduling by using Condor-G Matchmaking mechanism

An approach to grid scheduling by using Condor-G Matchmaking mechanism An approach to grid scheduling by using Condor-G Matchmaking mechanism E. Imamagic, B. Radic, D. Dobrenic University Computing Centre, University of Zagreb, Croatia {emir.imamagic, branimir.radic, dobrisa.dobrenic}@srce.hr

More information

A highly configurable and efficient simulator for job schedulers on supercomputers

A highly configurable and efficient simulator for job schedulers on supercomputers Mitglied der Helmholtz-Gemeinschaft A highly configurable and efficient simulator for job schedulers on supercomputers April 12, 2013 Carsten Karbach, Jülich Supercomputing Centre (JSC) Motivation Objective

More information

OPNET Network Simulator

OPNET Network Simulator Simulations and Tools for Telecommunications 521365S: OPNET Network Simulator Jarmo Prokkola Research team leader, M. Sc. (Tech.) VTT Technical Research Centre of Finland Kaitoväylä 1, Oulu P.O. Box 1100,

More information

RSA VIA LIFECYCLE AND GOVERNENCE: ROLE MANAGEMENT BEST PRACTICES

RSA VIA LIFECYCLE AND GOVERNENCE: ROLE MANAGEMENT BEST PRACTICES RSA VIA LIFECYCLE AND GOVERNENCE: ROLE MANAGEMENT BEST PRACTICES A practitioner s perspective on best practices for Role Management ABSTRACT This white paper provides an overview of the Role Management

More information

Application Performance Testing Basics

Application Performance Testing Basics Application Performance Testing Basics ABSTRACT Todays the web is playing a critical role in all the business domains such as entertainment, finance, healthcare etc. It is much important to ensure hassle-free

More information

Module 15: Network Structures

Module 15: Network Structures Module 15: Network Structures Background Topology Network Types Communication Communication Protocol Robustness Design Strategies 15.1 A Distributed System 15.2 Motivation Resource sharing sharing and

More information